The most expensive AI is the one that makes bad decisions. Data governance can prevent that, says Jan Štěpánovský, CETIN

June 25, 2025

CETIN is a major telecom infrastructure provider in Central and Eastern Europe. In this episode, CIO Jan Štěpánovský talks about the real value of AI in enterprise IT, the shift toward automation, and why data governance is essential for building AI strategies that work.

Read the podcast as an interview

(The interview was shortened and edited using ChatGPT)

Ivana Karhanová: You said during our prep that “sexy AI tools are overshadowing serious AI use in companies.” What do you mean?

Jan Štěpánovský: AI became a global buzzword over the past few years. Everyone sees something different in it. The easiest way to reach users is by building simple, polished tools—apps that suggest what to wear based on the weather or generate a meal plan. They’re everywhere, especially on app stores. Some are useful, others not so much. But they dominate how people think about AI.

Ivana Karhanová: And in business, that’s a problem?

Jan Štěpánovský: It shifts attention away from real use. I divide AI into four categories. First are those flashy tools we just talked about. Then there are AI agents—semi-smart automation tools that make work easier, especially in routine business processes. Third, assistants like voicebots and chatbots that help navigate internal documentation. And finally, advanced analytics and machine learning—serious, data-driven AI that delivers real business value.

Ivana Karhanová: But advanced analytics doesn’t sound nearly as exciting.

Jan Štěpánovský: No, and that’s the issue. What’s flashy gets the spotlight. But the real transformation will come from AI that improves how businesses work. In IT, I see AI integrating into roles like development, operations, testing, even security—not replacing people, but enhancing how they work.

Ivana Karhanová: How do you approach AI as a CIO?

Jan Štěpánovský: In two ways. First, how AI can help increase productivity and streamline processes. Second, how AI will reshape IT over the long term. I believe we’re heading toward multi-agent ecosystems. You’ll have one agent generating code, another checking it, another deploying it, others securing it. Together, they mirror the structure of modern IT teams.

Ivana Karhanová: Do you already have a strategy for that?

Jan Štěpánovský: We started working on our AI strategy about 18 months ago. Back then, it focused on chatbots and machine learning. It didn’t yet cover agents or full AI ecosystems. Now we’re updating it with those in mind.

Ivana Karhanová: So you’ve already had some real use cases?

Jan Štěpánovský: Yes. We’ve built chatbots for HR and IT, and implemented machine learning for network anomaly detection—something that would overwhelm a human. Now we’re focusing on connecting these pieces into a working AI ecosystem.

Ivana Karhanová: That brings us to data. Before AI, cloud was the big hype. What lesson did you take from that?

Jan Štěpánovský: Every tech hype has something valuable. The mistake is in blindly following it. A few years ago, many companies tried to move everything to the cloud—lift-and-shift style. But that often led to higher costs and no added value.
We faced this decision with our data warehouse. Rebuilding it in Azure using cloud-native components turned out cheaper than lift-and-shift and gave us better access to big data and machine learning tools.

Ivana Karhanová: So you built it differently—but not from scratch?

Jan Štěpánovský: Right. Architecturally, it’s new. But in terms of development, we reused and reconnected existing components. That made the whole project faster and more cost-effective.

Ivana Karhanová: What’s CETIN Ops?

Jan Štěpánovský: It’s our internal strategy combining DevOps, SecOps, and FinOps. We want to automate everything that can be automated, build security into everything we develop, and keep an eye on cost. It’s not just tools—it’s a mindset.

Ivana Karhanová: How far have you gotten?

Jan Štěpánovský: We manage around 300 systems. About 60 are critical, and half of those already use DevOps with automated builds, deployments, and tests—including security checks. The goal is to expand that coverage.

Ivana Karhanová: And what about infrastructure?

Jan Štěpánovský: We want to bring on-prem infrastructure closer to infrastructure-as-code. That’s a medium-term goal—very similar to what we’re doing in application development.

Ivana Karhanová: And security?

Jan Štěpánovský: Security needs to be a core part of development, not something added later. Developers should think about vulnerabilities from the start. Even an AI agent can become a risk if it’s not managed well.

Ivana Karhanová: Where does Adastra fit into this?

Jan Štěpánovský: You’ve helped us in two areas: DevOps and AI. We started working together on DevOps around 2019–2020. That system is now mature and ready for enterprise scale. With AI, we didn’t start with tools—we started with use cases. We evaluated each one seriously: will it bring value, how much will it cost, what are the risks? Only then did we move to implementation.

Ivana Karhanová: Let’s talk about data governance. With all your legacy and modern systems, how do you ensure data quality?

Jan Štěpánovský: We have about 300 systems, some over 25 years old. We started data governance in the new data warehouse: cleaning data, defining a data dictionary, checking quality. Now we’re extending that to legacy systems—which is much more complex.

Ivana Karhanová: Where would you place your maturity on a scale from 1 to 5?

Jan Štěpánovský: About 1.5. That doesn’t mean we’re behind—it means we’re pushing forward. Data governance, BI, and AI are tightly linked. Without understanding the meaning of your data, AI can’t work properly.

Ivana Karhanová: You’ve described a clear IT vision. What’s key to making it happen?

Jan Štěpánovský: People. You can have a great plan, a budget, and the right tools. But without the right people, nothing gets done. You need teams that understand and believe in the vision.

Ivana Karhanová: How hard is it to build such a team?

Jan Štěpánovský: Building from scratch is easy. The real challenge is evolving existing teams—full of great specialists who’ve been doing things a certain way. The key is to get them excited about the vision and let them lead the change.

Ivana Karhanová: And is it working?

Jan Štěpánovský: I think so. We’re seeing real progress, and we haven’t seen much turnover—so that’s a good sign.

Ivana Karhanová: Thanks for the conversation.

Jan Štěpánovský: Thank you for having me.